Model Prediction and Model loading

[charleswg]

I found that it doesn't load model for some reason. Maybe it's windows specific?

A quick fix change path checking in predict function from:

assert os.path.exists(self.params.modelPath)
to
assert os.path.exists(os.path.dirname(self.params.modelPath))

But when I try to predict after training,
ds_tf = deepsurv_tf.DeepSurvTF(params)
stats = ds_tf.train(train, valid)
ds_tf.predict(valid['x'])

The follow error occur:
--> 291 risk = sess.run([risk], feed_dict = {self.x : testXdata})

UnboundLocalError: local variable 'risk' referenced before assignment

[alexhallam]

Thanks for giving the code a shot!

Could you provide more information about [risk]?

From this line

--> 291 risk = sess.run([risk], feed_dict = {self.x : testXdata})

[charleswg]

Actually I was trying to load the trained model to predict the hazard ratio of incoming data.

But I took a step back and tried in-session prediction right after training.

Would you provide example on how to do prediction (hazard_ratio, log_risk etc) using saved model?

[alexhallam]

At the moment I am focusing on some other projects. I may not get to that in a timely manner. I tried to make the demo helpful, but of course I am sure there are many things (such as using a saved model) missing that would be helpful to many users. If other users are willing to push more ipython notebook examples that would be welcomed, as I do not plan on adding more examples any time soon. Also, freely submit pull requests as you try new things. Even if they are small things. I would be very accepting of ipython notebook demos. I am sure more examples could benefit someone else.